Establishing Rigorous Certification Standards: A Systematic Methodology for AI Safety-Critical Systems in Military Aviation

Abstract

This article presents a systematic methodology for developing a certification standard for AI safety-critical systems in military aviation, amalgamating military and civil airworthiness references. It involves thorough analysis to identify overlaps, contradictions, and specific needs for AI certification in this domain. The methodology entails incremental updates to a foundational certification framework, continuously integrating new references. An illustrative application to an ISO reference demonstrates the process of extracting AI certification requirements. Furthermore, systematically derived requirements from various ISO references exemplify the method’s efficacy. This systematic approach aims to consolidate pertinent information for establishing robust certification standards for AI safety-critical systems in military aviation. By integrating both military and civil aviation standards, it ensures comprehensive coverage of relevant criteria. The iterative nature of the methodology allows for ongoing refinement, reflecting evolving technologies and regulations. Ultimately, this approach fosters the creation of a standardized framework adaptable to the dynamic landscape of AI in military aviation. Through its meticulous analysis and synthesis of diverse references, the methodology strives to enhance the safety and reliability of AI systems deployed in critical military operations.

Keywords

• AI safety-critical systems
• artificial intelligence
• certification
• military aviation
• ISO references